Page 88 - Climate Change and Food Systems

climate change and food systems: global assessments and implications for food security and trade
degree of model consensus on particular variables, as well as for defining possible extreme outcomes. For a number of reasons, however, the range
in climate projections for each SRES does not represent a probability distribution of outcomes (Knutti, 2010); Masson and Knutti, 2011). The small number of participating GCM models are
not independent as they all contain similar laws of nature, they calibrate to and describe the same climate, and they may make the same simplifications in using parameters to represent certain natural processes. Many models, too, are variants of the same genealogical family.
Because of the differences in climate change results across SRESs and GCMs, the selection
of both scenario and climate models is a key decision that to a large extent predetermines results of economic assessments. A number
of considerations point to the advantages of
using climate outputs from more than one SRES and more than one GCM. The IPCC (2000) recommends that two or more individual scenarios (that is, without combining or splicing them) be drawn from more than one SRES family because of the uncertainties associated with the likelihood of any scenario. Many of the economic assessments
surveyed here analyse SRES A1B, which is a mid- to upper-range emission scenario that describes “business as usual,” with no mitigation, and SRES A2 and A1F to describe increasingly pessimistic future emission levels. SRES B1 describes an optimistic scenario with emission stabilization.
The selection of GCMs is constrained by whether they report the required climate variables at the spatial and temporal resolutions needed
by the impact study. Within the subset of suitable GCMs, it is efficient to choose the model and SRES combinations that represent the widest range in output values for the variables of
interest. However, these ranges typically vary across regions and by variable. For example,
the same GCM/SRES combination may provide the most extreme outcomes for temperatures in humid tropical zones, but provide little range in precipitation variables compared to other GCM/ SRES combinations. Model performance, in terms of GCM’s ability to simulate present-day climate, seems a reasonable selection criterion, but one that can be difficult to actually apply. Models’ comparative performances vary across different climate variables and, ultimately, it cannot be known which models predict the future best
figure 2
Range of carbon dioxide emissions from SRES scenarios
Source: IPCC (2000)
68